Faster inverse gamma and factorial for the WP 34S

[BarryMead]

(02-08-2015 09:44 PM)rprosperi Wrote:  Sure, this is certainly correct and admirable when the balance of effort and resources are justified by the need. I'm just asking about the need, as I don't see the practical applications driving this need.

Not everyone "Needs" to put the calculator into Double Precision mode, but when one does s/he expects the results to be more accurate. It is very rare indeed that one NEEDS 34 digits of accuracy, but when you perform an operation in Double Precision mode wouldn't you expect the 34 digit result of any function to be reasonably correct? My point is why ask a person why he NEEDS something? Isn't it better to just assume that he has a valid need, and deliver the results he expects. The alternative would be to inject arbitrary errors assuming that his need for accuracy is not valid. (In this case one would usually add a disclaimer explaining the accuracy limits of the algorithm so that people don't rely on it.) There are rarely "PERFECT" algorithms, so researchers are always STRIVING to improve them balancing factors such as program size, computation speed, and accuracy. Some find these challenges interesting, competitive, and fun, and others may not.